The Surgical Treatment of Andersson Lesions Associated With Ankylosing Spondylitis

Materials and Methods

Eight men with Andersson lesions associated with ankylosing spondylitis who underwent surgical treatment in our department from 2004 to 2010 were included in this study. Basic patient information such as body height, body weight, and follow-up time was recorded. A detailed history was recorded and a physical examination was performed for each patient preoperatively. Human leukocyte antigen B27 (HLA-B27) analysis, erythrocyte sedimentation rate, and C-reactive protein level were also obtained preoperatively.

Standing anteroposterior (AP) and lateral radiographs were obtained preoperatively, 10 days postoperatively, and at follow-up. On radiographs, several parameters were studied: level of the Andersson lesion, local kyphosis, and global kyphosis. Local kyphosis in the lesion region was estimated as the angle between the lines drawn from the upper endplate of the cranial vertebral body and the inferior endplate of the caudal vertebral body. ^7,8 Global kyphosis was estimated as the Cobb angle of the thoracolumbar or the whole spine (Figure ).

Figure 1:. Radiograph showing global kyphosis: the Cobb angle of the whole spine (A). Radiograph showing local kyphosis: the angle between the lines drawn from the upper endplate of the cranial vertebral body and the inferior endplate of the caudal vertebral body (B).

Magnetic resonance imaging (MRI) and computed tomography (CT) were performed preoperatively for all patients to detect spinal cord compression and identify the conditions of the anterior elements (including intervertebral disk and vertebral body) and the posterior elements (including vertebral plate, ligamentum flava, interspinal, and supraspinal ligament).

Two surgical techniques were used in this study: (1) posterior instrumentation with lesion curettage and anterior bone graft; and (2) posterior instrumentation without lesion curettage and anterior bone graft. Because there are no established instructions for the surgical treatment of Andersson lesions, although we performed all surgeries from a posterior approach, we alternated techniques for each patient. In addition, posterolateral autograft was supplemented to posterior instrumentation with or without anterior bone graft. Biopsy was performed in several patients, and other detailed surgical information, such as operation time and blood loss, was recorded.

Results

Eight Andersson lesions, all pseudoarthrosis, were observed in the 8 patients. All 8 Andersson lesions occurred in the intervertebral space, mainly at the thoracolumbar region.

All 8 patients were men, with a mean age of 45.6 years (range, 28–57 years). Mean height was 171.0 cm (range, 169.0–175.5 cm), and mean weight was 75.1 kg (range, 67–85 kg). The period between the onset of persistent back pain and surgery ranged from 3 to 8 years. Mean follow-up was 3.75 years (range, 2–6 years) (Table ).

Table 1. Patient Data

All 8 patients reported persistent back pain for years that had progressed in the months before the operation. Six of the 8 patients reported a history of trauma at the onset of the persistent back pain, but no patient reported trauma at the time of progression of the back pain. The period between the onset and progression of the back pain ranged from 3 to 8 years. Except in 1 patient with decreased limb strength, no neural deficit was detected in any patient.

For the 8 patients, local kyphosis was 10.9° preoperatively (range, 7.0°–16.5°) and 6.75° postoperatively (range, 1.0°–12.0°). Global kyphosis was observed in 4 patients preoperatively, with a mean value of 48.3° (Table ).

Table 2. Radiological Data

For all 8 Andersson lesions, destruction of anterior elements and fracture of posterior elements were observed on MRI. In 1 patient, canal encroachment was observed on both MRI and CT.

Mean operation time was 4.1 hours (range, 3–6 hours). Mean blood loss was 1150 mL (range, 810–1500 mL).

Three patients underwent surgery with posterior instrumentation without lesion curettage and anterior bone graft, while 5 patients underwent surgery with both posterior instrumentation and lesion anterior curettage and bone graft. Posterolateral autograft was performed to supplement the posterior instrumentation with or without anterior lesion curettage and bone graft. No neurological or infectious complication occurred. At final follow-up, no evidence of nonunion at the level of pseudoarthrosis was observed.

Biopsy was performed in 4 patients undergoing anterior lesion curettage and bone graft and in 1 patient undergoing surgery without lesion curettage. Histological study showed the lesions were infiltrated by plasma cells and lymphocytes. No infectious inflammation was observed.

Discussion

Andersson lesions are a common complication associated with ankylosing spondylitis, presenting as the development of localized vertebral or diskovertebral lesions of the spine. ^1,6

Two possible etiologies, inflammatory and mechanic, are described in the literature. ^2–7,9–15 Dihlmann and Delling ² divided Andersson lesions into 2 main types: inflammatory and noninflammatory; however, this does not apply to all cases of Andersson lesions. Therefore, although still a topic of debate, a common opinion is that both inflammatory and mechanic factors play a role in the origin and development of Andersson lesions. ^{6,9,12,13,15–17}

In this study, we found a history of trauma at the time of onset of persistent back pain in 6 of our 8 patients. We believe that trauma is the possible origin of Andersson lesions in our study; however, we did not obtain radiological records from the onset of back pain, and we found no history of trauma at the time of progression of back pain.

Bron et al ⁶ divided Andersson lesions into 3 groups according to radiological characteristics: localized lesions, extensive lesions with fractured posterior elements, and extensive lesions without fractured posterior elements. In our study, all the Andersson lesions were extensive lesions. Among the 8 Andersson lesions, rupture of posterior ligaments was observed in 7 as interspinous edema on MRI, and fracture of the spinous process was observed in 1.

Two theories about the reasons for destruction of the vertebral body and intervertebral disk are more convincing than others. The first is that the nuclear material that herniates through the endplate into the vertebral body causes inflammatory destruction because of its antigenicity. ⁶ The other, more common theory is that local nonunion of the ankylosed spine results from increased spinal inflammation and an excessive degree of mobility caused by inflammatory destruction. ^5,9,12 Bron et al ⁶ theorizes a “common pathway” to interpret the development of pseudoarthrosis that emphasizes the cooperation of inflammatory and mechanic (instability) factors.

There are also 2 possible causes for the rupture of posterior elements: primary trauma or stress fracture secondary to anterior destruction. According to Bron et al, ⁶ the type with fractured posterior elements was mainly caused by primary trauma. Several early reports on Andersson lesions without fractured posterior elements proved the more important role of instability. ^10,13,14

According to the radiological parameters obtained in this study, no obvious local kyphosis occurred in the lesion region, even in the lesion with obvious anterior destruction, so there was no tension on the posterior elements to cause secondary fracture. Therefore, we have reason to believe that the destruction of anterior elements and the fracture of posterior elements occurred at the same time as the original trauma. The inflammatory reaction was trigged to cause vertebral and intervertebral destruction, while instability caused by the fracture of posterior elements hindered reunion.

Regarding the management of Andersson lesions, several researchers report that conservative treatment could be effective. ^2,4,13 One report describes the spontaneous relief of back pain in patients with Andersson lesions. ² However, Bron et al ⁶ pointed out that conservative management is less efficient at the more mobile cervical and thoracolumbar levels. Conservative management is also not suitable for pseudoarthrosis, because without solid fixation gained from surgery, minimal persistent motion at the Andersson lesion might hinder fracture healing and union. Neither brace immobilization nor Halo jacket immobilization could eradicate this minimal persistent motion. ^12,15

Nonsteroidal anti-inflammatory drugs and antitumor necrosis factor-a are commonly used in the conservative management of Andersson lesions; however, Bron et al ⁶ proposed that there is no evidence that treatment with these drugs is beneficial in the treatment of symptomatic Andersson lesions, although they are effective in the treatment of ankylosing spondylitis.

Regarding the debate on surgical treatment, the main argument is focused on the necessity of anterior bone graft and fusion. Chang et al ^7,8 reported that anterior fusion was not necessary because of the excellent reunion ability of ankylosing spondylitis; however, Chen et al ¹⁸ reported the effectiveness of surgical treatment with both posterior fixation and anterior fusion. Because there are no established instructions for the surgical treatment of Andersson lesions, although we performed all surgeries from a posterior approach, we alternated techniques for each patient. At first, we believed it was necessary to perform lesion curettage and bone graft. However, in 2007, we treated a patient who could not tolerant lesion curettage and anterior bone graft, so we only used posterior instrumentation. At 1-year follow-up, union of the pseudoarthrosis was achieved (Figure ). In 2008, we performed surgeries with posterior instrumentation on 2 other patients. In 2009, for a patient with obvious vertebral body destruction and neural deficit, we performed posterior decompression and anterior bone graft but not excessive lesion curettage (Figure ). The outcomes of the 8 patients treated in our department were good. No evidence of nonunion was observed in any patient at last follow-up.

Figure 2:. Radiographs of a patient who underwent surgical treatment without lesion curettage and anterior bone draft. An Andersson lesion occurred at T12/L1, and there was no nonunion at final follow-up (A). MRI showing destruction of intervertebral disk and rupture of the interspinous (B).

Figure 3:. Radiographs of a patient with obvious anterior destruction. Although there was an obvious sagittal imbalance, the local kyphosis was only 16.8°. The global kyphosis was mainly generated in the proximal thoracic region (A). CT and MRI showing obvious anterior destruction (B). Radiographs showing pseudoarthrosis union at final follow-up (C).

Based on these results, we believe that solid immobilization achieved by posterior instrumentation is the most significant treatment for Andersson lesions, which also proves the significant role of instability in the development of Andersson lesions. In addition, in the Smith-Peterson osteotomy, the intervertebral disk was always ruptured with nuclear material exposed, ¹⁹ which is similar to 1 hypothesis of Andersson lesions. ^5,6,9 However, there are no reports of Andersson lesions occurring at the segment of Smith-Peterson osteotomy.

Conclusion

For patients with Andersson lesions without obvious vertebral body destruction requiring reconstruction, surgical treatment with posterior instrumentation supplemented by posterolateral autograft was effective. Lesion curettage and anterior bone graft was notnecessary. Solidimmobilization achieved by posterior instrumentation should be the focus of the treatment of Andersson lesions associated with ankylosing spondylitis.